Applied Geography 120 (2020) 102228

Contents lists available at ScienceDirect

Applied Geography
journal homepage: http://www.elsevier.com/locate/apgeog

Simulating the effects of land urbanization on regional meteorology and air

quality in Yangtze River Delta, China
Tan Chen a, b, Manchun Li a, b, *, Lina Luo c, Shulin Deng a, b, Rui Zhou d, Dong Chen a, b
a
School of Geography and Ocean Science, Nanjing University, Nanjing, 210023, China
b
Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology, Nanjing University, Nanjing, 210023, China
c
Department of Civil and Environmental Engineering, Rice University, Houston, TX, USA
d
School of Environmental and Geographical Sciences, Shanghai Normal University, Shanghai, 200234, China

A R T I C L E I N F O A B S T R A C T

Keywords: Land urbanization caused by human activities is increasingly converting natural land surfaces to artificial ones
Land urbanization and modifying land surface parameters. This can lead to changes in weather conditions and further effects on air
Air quality simulation pollutants. This study examines the influence of land urbanization on meteorology and air quality in the Yangtze
Urban sustainability
River Delta (YRD) by simulating the impacts of land-use/land cover (LULC) changes on meteorology and air
Yangtze River Delta
quality. Our results show that the temperature at 2m increases by 0.2–0.7 � C, and the planetary boundary layer
height exhibits a maximum growth during the daytime of 116.74 m after updating LULC. The average reduction
of 10-m wind speed is 0.29 m/s, and the maximum decrease is 0.52 m/s, which occurs in the daytime due to
increased surface resistance. Furthermore, PM2.5, NO2, SO2, and ozone are sensitive to these meteorological
changes in the WRF-CMAQ model. Additionally, land urbanization results in a decrease in the PM2.5 concen­
tration in most urban areas for the same emission source. However, land urbanization aggravates photochemical
smog pollution, which is dominated by O3. The findings of the study can be used as a reference for policymakers
and urban planners to help implement effective air pollution prevention and control measures and ensure the
rational use of land resources.

1. Introduction (2018) analyzed the relationship between urbanization and fine par­
ticulate matter concentrations using the Environmental Kuznets Curve
Since China’s reform and opening up, urban land expansion has hypothesis. The effects of urbanization on CO2 emissions have been
changed many natural land surfaces into artificial ones due to the rapid discussed from the perspectives of population and land urbanization
development of urbanization. The increasingly serious environmental using an extended Stochastic Impacts by Regression on Population,
problems generated by land urbanization (e.g., climate change, urban Affluence, and Technology model (Zhang et al., 2018). Furthermore,
heat islands, air pollution) have attracted the attention of environmental land-use regression models have been widely adopted to estimate air
managers and researchers. However, research into the effect of land quality based on land-use cover data (Dirgawati et al., 2015; Haber­
urbanization on regional meteorology and air quality is still a contro­ mann, Billger, & Haeger-Eugensson, 2015; Hoek et al., 2008). Land
versial but key topic. urbanization plays an important role in China’s urbanization process
Many empirical studies have described the relationship between because both the land-use properties and property rights attributes have
urbanization and the atmospheric environment from different perspec­ changed (Lin et al., 2017; Xu & Zhang, 2016). Therefore, this study
tives and using different methods (Xu, Dong, & Yang, 2018; Zhang et al., analyzes how land urbanization affects regional meteorology and air
2015, 2018). For example, previous studies have focused on different quality to provide a better explanation of China’s current urbanization
pollutants, including fine particulate matter (Du et al., 2018), carbon and air quality trends.
emissions, and gaseous pollutants such as CO2, NOX, SO2, and O3 (Lin, Although there are previous literature on the impact of land urban­
Wang, Marinova, Zhao, & Hong, 2017; Noor, Hasim, & Yusof, 2018; ization on air quality aforementioned, debates of these researches and
Yang, Wang, Zhang, Zhan, & Li, 2016). Wu, Zheng, Feng, Xie, and Jing key research questions accurately measure air quality. A wide variety of

* Corresponding author. School of Geography and Ocean Science, Nanjing University, Nanjing, 210023, China.
E-mail address: limanchun@nju.edu.cn (M. Li).

https://doi.org/10.1016/j.apgeog.2020.102228
Received 3 February 2020; Received in revised form 16 April 2020; Accepted 8 May 2020
Available online 1 June 2020
0143-6228/© 2020 Elsevier Ltd. All rights reserved.
T. Chen et al. Applied Geography 120 (2020) 102228

methods have been reported for capturing or predicting regional air CMAQv5.1) was adopted to simulate the impacts of land urbanization on
quality. The traditional means of field observations and laboratory regional meteorology and air quality over the YRD region. The WRF
simulations of atmospheric pollutants are highly reliable and practical. model provides meteorological field elements for CMAQ simulation, and
However, due to the limitations of observation sites, observation time, the CMAQ model serves as the core model for air quality simulation.
simulation conditions, and other factors, it is impossible to obtain Operation of the WRF-CMAQ model adopts a one-way triple nesting
pollutant concentration distributions with high temporal resolution and mode to simulate the region with a Lambert projection. The specific
wide coverage (Steinfeld, 1998). Compared with ground-based moni­ distribution is shown in Fig. 1. The resolution of the outermost simula­
toring, the remote sensing inversion method has the advantage of tion region is 36 km � 36 km, covering mainland China (Domain 1). The
quickly obtaining broad spatial coverage data (Engel-Cox, Holloman, resolution of the middle layer simulation region is 12 km � 12 km and
Coutant, & Hoff, 2004; Hoff & Christopher, 2009). However, satellite covers eastern China (Domain 2). The resolution of the inner layer
data is typically more suitable for large-scale research areas with a simulation region is 4 km � 4 km, covering the YRD region (Domain 3).
coarse spatial resolution (Wang & Christopher, 2003; Fioletov, Mclin­ In the WRF model, the simulation region is divided into 35 vertical
den, Krotkov, Moran, & Yang, 2011). Nevertheless, chemical transport layers from the ground to the upper air using the sigma coordinate
models still require improvement due to uncertainties in emission in­ system. The top height of the layer is 5000 mb, including all regions of
ventories and the chemical and dynamical mechanisms of the atmo­ the troposphere (the top pressure of the troposphere is approximately
sphere (Karambelas, Holloway, Kiesewetter, & Heyes, 2018; Pisoni 10,000 mb). The initial and boundary conditions are derived from the
et al., 2018). global reanalysis data of NCEP (National Centers for Environmental
The Yangtze River Delta (YRD) region, one of the most polluted re­ Prediction), with a horizontal resolution of 1� � 1� and a temporal
gions in China, has experienced rapid economic development and land resolution of 6 h. The following physical options are adopted in the main
urbanization in recent decades. Rapid economic development in this scheme of the one-way nesting simulation: the Pleim-Xiu land-surface
region has been accompanied by a huge consumption of energy, which model (Pleim & Xiu, 2003), the Grell-3 cumulus scheme (Grell &
has led to air pollution (Liang, Cai, & Ma, 2019; Ma, Cai, Cai, & Dong, Devenyi, 2002), the RRTM radiation mechanism (Mlawer, Taubman,
2019). Therefore, the YRD region has become a key research region. Li Brown, Iacono, & Clough, 1997), and Purdue-Lin microphysics (Lin,
et al. (2011) studied air quality and aerosol emissions in the YRD region Farley, & Orville, 1983).
using the MM5-CMAQ (Community Multiscale Air Quality) model and In the CMAQ simulations, the horizontal simulation region and res­
revealed that ozone and haze had become an important air quality issue. olution are approximately the same as those of the WRF model; how­
Wang and Christopher (2003) used the Weather Research and Fore­ ever, to ensure the accuracy of the boundary meteorological conditions,
casting model coupled with Chemistry (WRF-Chem) to examine the in­ the horizontal region of the air quality model is slightly smaller than that
fluence of urbanization on surface ozone in YRD, reporting that of the WRF model. In the vertical direction, 14 sigma layers are adopted
urbanization increased the surface ozone concentration during both the with varying thickness from the surface to 5000 mb. Moreover, CB05
daytime and nighttime by 2.9–4.2% and 4.7–8.5%, respectively. More was selected as the gas-phase mechanism (Yarwood, Rao, Yocke, &
recently, Xie et al. (2016) investigated the spatial distribution of Whitten, 2005), AERO6 as the aerosol module (Prakash, Jeremiah, Greg,
anthropogenic heat emissions in the YRD region and their impact on & Eladio, 2014), ISORROPIA II as the thermodynamic model (Nenes,
regional meteorology and air quality using the WRF-Chem model. Pandis, & Pilinis, 1998), and ACM_AE6 as the cloud module. The
The urbanization process in YRD is still developing rapidly, yet emission inventory used in this study is the China multi-scale Emission
research on the impacts of land urbanization on regional meteorology Inventory (MEIC) developed by Tsinghua University. It includes pro­
and air quality by high precision method in this region remains insuf­ vincial data summarized for five sectors: power, industry, residential,
ficient. Land-use/land cover changes (LUCC) due to urbanization also transportation, and agriculture. The simulation period is from 0000 LST
influence the regional air quality, yet these underlying surface changes 01 January to 0000 LST 01 February 2016.
have also received minimal research attention. Many previous studies
have focused on land-use patterns or statistical analyses such as land-use 2.2. Land-use map in simulation design
regression models. However, they have not considered the chemical
transport mechanism, and neglected the transfer traces and process of IGBP (International Geosphere–Biosphere Programme) land classi­
atmospheric pollutants. To fill the gaps above, we combine the factors fication has been used instead of USGS (United States Geological Survey)
influencing LUCC (utilizing the latest YRD land-use data from 2015) as the default for WRFv3.9. However, this map is generated from 30
with an air quality simulation model (the WRF-CMAQ Model) to study arcsec (approximately 1 km) MODIS (Moderate Resolution Imaging
the impacts of urbanization-related surface changes on meteorology and Spectroradiometer) data for 2001, with 20 land cover/land-use classi­
regional air quality in YRD by hour interval. The findings of the study fications. Therefore, it is relatively old compared to our simulation
can be used as a reference to air quality management and decision- period of January 2016. As such, we use new land-use map data from the
making of air pollution control and provides reliable support for envi­ Resource and Environment Data Cloud Platform of the Chinese Academy
ronmental planning towards urban sustainability. of Sciences (http://www.resdc.cn). The new land-use map is generated
The remainder of this paper is organized as follows: Section 2 pre­ for 2015 with a horizontal resolution of 1 km and reclassified for the
sents the methodology and data. The methods include model descrip­ same 20 land-use categories as IGBP-MODIS (termed the LULC_NEW
tion, model setup, and scenario design. For the data, we mainly scenario). Then, the default land-use scenario is defined as a base
introduced the data source and attributed applied in the model. Section simulation (termed the LULC_MODIS scenario). Fig. 2 compares the
3 shows the model evaluation. For Section 4, Section 4.1 presents the LULC_MODIS map with 1-km resolution to the new 1-km resolution
results about the impacts of land urbanization on meteorology. Section LULC_NEW map created in this study for WRF-CMAQ simulation, which
4.2 reveals the impacts of land urbanization on pollutant concentration. shows that the YRD region is experiencing severe land urbanization.
The key findings and future research outline will be summarized in
Section 6. 3. Model evaluation

2. Methodology and data 3.1. Meteorology

2.1. Model description and setup The simulation performance of the WRF model was evaluated by
comparing the results of the LULC_NEW scenario with hourly data from
In this study, a coupled model WRF-CMAQ (WRFv3.9 and ground meteorological stations of NOAA (National Oceanic and

2
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 1. (a) Simulation study area and (b) distribution of ground observation stations (red stars are urban meteorological stations and green circles are urban air
quality stations). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2. Two scenarios used in the simulations: 2001 land-use map from MODIS (left) and 2015 land-use map reclassified to match the categories of IGBP-
MODIS (right).

Atmospheric Administration). We chose typical statistical metrics Table 1. The correlation coefficients (R) of the T2 simulation results in
including the correlation coefficient (R), mean bias (MB), root mean Shanghai, Nanjing, and Hangzhou are 0.98, 0.99, and 0.99, respectively,
square error (RMSE), normalized mean bias (NMB), and normalized indicating that the WRF simulation results capture the T2 variation
mean error (NME), following the evaluation protocol of Zheng et al. characteristics. The MBs are 0.03, 0.26, and 0.15, and the NMBs are
(2015). As shown in Fig. 3, the temperature at 2 m and wind speed at 10 0.61%, 6.91%, and 2.75%, which agree well with the observations.
m simulated by the WRF model better capture the actual temperature For WS10, the R values are lower than for the T2 simulation result.
change process, where the blue line is the simulated value time series WS10 is slightly underpredicted with an NMB of 13.34% and an MB of
curve, and the red points are the hourly observation values of the three 0.33 m/s for Hangzhou. Overpredicted or underpredicted results are
urban surface meteorological stations. probably due to the influence of local underlying surface characteristics
The statistical metrics for meteorological factors, including temper­ in urban areas or the parameterization scheme (Zhang, Gao, Wang, &
ature at 2 m (T2) and wind speed at 10 m (WS10), are presented in Chen, 2010). In general, the simulated meteorological variables show

3
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 3. Hourly simulation results of 2-m temperature and 10-m wind speed compared with observation data in Shanghai, Nanjing, and Hangzhou city meteorological
stations for January 2016.

Table 1
Performance statistics of the WRF simulation.
Variables Sitesa Statistical metrics

MSb MOb MB RMSE R NMB (%) NME (%)

T2 ( C)
�
SH 4.96 4.99 0.03 0.76 0.98 0.61 11.34
NJ 4.09 3.83 0.26 0.76 0.99 6.91 15.18
HZ 5.70 5.55 0.15 0.67 0.99 2.75 9.67

WS10 (m/s) SH 3.79 3.75 0.04 0.77 0.92 1.07 12.31

NJ 2.56 2.50 0.06 0.82 0.84 2.27 26.51
HZ 2.16 2.49 0.33 0.83 0.86 13.34 26.96
a
Abbreviated city names: SH ¼ Shanghai, NJ ¼ Nanjing, and HZ ¼ Hangzhou.
b
MS and MO represent the mean simulation and observation values, respectively.

good agreement with the observed values. from 24.56% to 41.19%. The simulated pollutant variables generally
show good agreement with the observed values.
3.2. Air quality
4. Results and discussion
Values of NMB, NME, R, normalized mean bias factor (NMFB), and
4.1. Impacts of land urbanization on meteorology
normalized mean error factor (NMEF) are used to evaluate the CMAQ
performance in this study. Time-series comparisons for PM2.5 and O3
4.1.1. 2-m temperature
concentrations are shown in Fig. 4, and statistical metrics are calculated
Fig. 5 shows the simulated difference of average monthly tempera­
in Table 2. These results show that the CMAQ model effectively re­
ture at 2 m (T2) between LULC_ NEW and LULC_MODIS scenarios in
produces changes of PM2.5 and O3 concentration in the YRD region.
January 2016. The difference range of T2 is between 1 and 1 � C, but
As shown in Table 2, the R values of PM2.5 in Shanghai, Nanjing, and
predominantly between 0.5 and 0.5 � C. After updating the land-use/
Hangzhou are 0.72, 077, and 0.69. NMB ranges from 16.24% to
coverage data, T2 exhibits spatially and temporally heterogeneous
22.69%, and NME ranges from 20.96% to 34.86%. The PM2.5 concen­
changes, and the main change occurs in urban areas with obvious land
trations are overestimated in Nanjing city and underestimated in
urbanization. Besides, there is a difference in the T2 change between
Shanghai and Hangzhou city. Previous studies suggest that the air
daytime and nighttime. Specifically, the average T2 decreases by
quality model can meet the simulation performance standard of air
0.2–0.6 � C at both 20:00 Local Standard Time (LST) and 02:00 LST in the
pollutants when NMBF and NMEF are less than or equal to �60% and
night. On the contrary, the T2 mean value does not change significantly
þ75%, respectively (Boylan & Russell, 2006; EPA, 2007). The LULC_­
at 08:00 (daytime) but exhibits a significant increase of 0.2 � C–0.7 � C at
NEW case agrees well with observations of PM2.5 concentrations with
14:00 LST. The rise of T2 is more evident in the daytime, which may be
NMBF values from 19.39% to 22.69% and NMEF values from 25.02%
related to the increased impervious surfaces of cement pavement and
to 37.54%. For O3, the simulation results capture the O3 variation
other high-rise buildings due to land urbanization. An increase in the
characteristics well, with NMBF from 11.64% to 14.91% and NMEF

4
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 4. Hourly simulation results of PM2.5 and O3 compared with observation data from Shanghai, Nanjing, and Hangzhou city AQI stations for January 2016.

Table 2
Performance statistics of the CMAQ simulation.
Variables Sitesa Statistical metrics

MSb MOb R NMB (%) NME (%) NMBF (%) NMEF (%)

PM2.5 SH 58.47 69.81 0.72 16.24 20.96 19.39 25.02

NJ 97.88 79.77 0.77 22.69 27.62 22.69 27.62
HZ 72.98 78.59 0.69 7.14 34.86 7.68 37.54

O3 SH 44.63 51.28 0.65 12.97 21.37 14.91 24.56

NJ 32.95 38.39 0.62 14.17 26.98 16.51 31.44
HZ 24.32 27.15 0.82 10.43 36.89 11.64 41.19
a
Abbreviated city names: SH ¼ Shanghai, NJ ¼ Nanjing, and HZ ¼ Hangzhou. Each city contains multiple ground monitoring stations.
b
MS and MO represent the mean simulation and observation values, respectively.

heat capacity and conduction efficiency of solar radiation in urban areas data, the changes of PBLH are similar to the diurnal variations and
leads to more heat storage during the day and heat release at night. spatial distributions of T2. Due to the evolution of thermal flux caused
by the turn of T2, PBLH changes substantially at noon but minimally
4.1.2. 10-m wind speed before dawn. PBLH increases significantly in the daytime in urban areas
Fig. 6 shows the simulation difference of monthly average wind and typically decreases at night. Specifically, the average PBLH drops
speed at 10 m (WS10) in January 2016 between LULC_NEW and 10.23 m–157.86 m at 20:00 LST and 02:00 LST. However, the PBLH
LULC_MODIS scenarios. After updating the LULC, the surface roughness value at 8:00 LST in the daytime does not change significantly. From the
of the YRD region exhibits significant changes, which contributes to perspective of spatial heterogeneity, PBLH changes predominantly occur
further increased surface resistance in urban areas and weaker wind in urban areas because they store more heat. For example, PBLH in­
speeds. In YRD, the average reduction of WS10 is 0.29 m/s, and the creases considerably in urban areas in the YRD region at 14:00 LST, with
maximum decrease is 0.52 m/s, which occurs at 14:00 LST. The varia­ a maximum increase of 116.74 m.
tion of wind speed also displays regional characteristics and temporal
heterogeneity. In urban areas, especially cities far from the coastal zone,
the wind speed shows more obvious weakening, which is closely related 4.2. Impacts of land urbanization on pollutant concentration
to the process of land urbanization. During land urbanization, WS10
decreases sharply in urban areas due to the increased roughness. 4.2.1. Fine particulate matter
Nighttime changes of WS10 are similar to those during the day, but the In this study, PM2.5 is chosen as the indicator to analyze the influence
reduction is significantly smaller. of land urbanization on particulate matter pollution (Fig. 8). After
updating LULC, the PM2.5 concentration change in the YRD region
4.1.3. Planetary boundary layer height shows a robust diurnal trend. During the daytime, there is a profound
PBLH is a crucial factor that affects near-surface pollutants during reduction in urban areas ranging from 1.04 to 6.53 μg/m3. At night,
vertical mixing and dilution (Su, Li, & Kahn, 2018). The spatial distri­ some areas such as the southern part of Shanghai, the northern part of
bution differences in the PBLH simulation results between LULC_NEW Hangzhou, and other urban areas show a smaller range of increase due
and LULC_MODIS scenarios are shown in Fig. 7. After updating the LULC to the high background concentration of PM2.5 accumulated during the
day. The maximum growth is 6.17 μg/m3 at 02:00 LST and 20:00 LST.

5
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 5. Spatial distribution differences of monthly-averaged T2 (� C) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00 LST, c)
14:00 LST, and d) 20:00 LST.

The spatial heterogeneity in cities is also reflected by the decrease of and becomes the main factor influencing PM2.5 reduction. Moreover, the
PM2.5 concentration in Nanjing between day and night and the apparent spatial distribution of PM2.5 reduction at 14:00 overlaps with the
diurnal differences in PM2.5 concentration changes in Shanghai and increased PBLH to some extent. During the night, due to the high PM2.5
Hangzhou. concentration in the background and the minimal vertical diffusion of
Also, meteorological factors have some influence on the diurnal PBLH, PM2.5 still shows a downward trend. Furthermore, the decrease of
variation of PM2.5 concentration. In the daytime, PBLH increases sharply PM2.5 is mainly concentrated in areas where WS10 also decreases.

6
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 6. Spatial distribution differences of monthly-averaged WS10 (m/s) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00
LST, c) 14:00 LST, and d) 20:00 LST.

Particularly in the afternoon, the statistical result at 14:00 shows that which is enhanced by the increased PBLH.
the spatial pattern of PM2.5 is approximately consistent with the
decrease of WS10. This phenomenon indicates that the horizontal 4.2.2. Gaseous pollutants
transport of PM2.5 is weakened in the area where the wind speed de­ As gaseous pollutants, we selected conventional monitoring pollut­
creases. It also proves that the decline of PM2.5 during the day is pre­ ants of the national control station of China. For a convenient compar­
dominantly caused by the vertical diffusion of particle concentration, ison with ground data in China, the units of gaseous pollutants were

7
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 7. Spatial distribution differences of monthly-averaged PBLH (m) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00 LST,
c) 14:00 LST, and d) 20:00 LST.

converted from ppb to μg/m3 for the model simulation. The increase and extent of O3 in the afternoon is higher than that in the
Precursors and solar radiation are the main factors affecting the morning, with a maximum of 2.19 μg/m3. The growth during the day is
formation of O3. Fig. 9 shows the difference in monthly average O3 more concentrated in central urban areas, whereas non-urban central
concentration for January 2016 in LULC_NEW and LULC_MODIS sce­ areas exhibit an increase and a slight decline at night. This change of O3
narios. After the LULC data is updated, the O3 concentration in urban concentration may be related to the change of meteorological parame­
YRD areas increases over a wide range and exhibits a diurnal difference. ters caused by land urbanization. Although the PBLH increases during

8
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 8. Spatial distribution differences of monthly-averaged PM2.5 (μg/m3) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00
LST, c) 14:00 LST, and d) 20:00 LST.

the day, the O3 concentration also increases. Furthermore, locations process; it is generally produced during the day, disappears in the eve­
where wind speed decreases are consistent with those where O3 con­ ning, and peaks at approximately 14:00 in the afternoon. Due to the
centration rises. It is notable that, although the O3 concentration change weak solar radiation in winter, it is more affected by NOx. During the
has a similar spatial distribution to that of PM2.5, the trend is entirely day, O3 is negatively correlated with NO2 because NO2 photodissocia­
opposite, resulting in a negative correlation during the study period. As tion provides atomic oxygen. At night, due to the oxidation of NO, the
the main component of photochemical smog, O3 exhibits a cyclic concentration of NO2 begins to increase, which indicates that NO2 and

9
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 9. Spatial distribution differences of monthly-averaged O3 (μg/m3) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00
LST, c) 14:00 LST, and d) 20:00 LST.

O3 at night will sometimes show a positive correlation. Furthermore, precursors NO2 and SO2, respectively, which suggest that they may be
NO2 is the precursor of PM2.5, and the trends of the two are approxi­ affected by changes in meteorological conditions. In general, the evo­
mately equal (Chen et al., 2018; Ito, Ogino, Nagaoka, & Takemoto, lution of PBLH has a particular influence on NO2 and SO2. The change of
2018). Therefore, O3 surface concentration has a negative correlation NO2 ranges from 2.25 to 3.25 μg/m3, and the transformation of SO2
with the PM2.5 level. varieties from 4.39 to 8.51 μg/m3. When PBLH increases, NO2 and SO2
Figs. 10 and 11 show the simulated daytime variations of the decrease, which is similar to the trend of PM2.5. This is because the

10
T. Chen et al. Applied Geography 120 (2020) 102228

increase of PBLH increases the vertical diffusion of gaseous pollutants. 14:00, the spatial distribution of PM2.5, NO2, and SO2 changes are
The effect of wind speed on SO2 and NO2 during the day is positively approximately equal. The spatial distribution of PM2.5 changes under
correlated, and the spatial distribution of the changes is approximately the joint influence of SO2 and NO2 are shown in Fig. 8. At 02:00 and
equal; i.e., a lower wind speed reduces the horizontal transport of 20:00 LST, the concentration distributions of SO2 and PM2.5 are more
gaseous pollutants. As precursors of PM2.5, changes in the concentration similar than those of NO2 and PM2.5, indicating that the components of
of SO2 and NO2 together affect the concentration change of PM2.5. At PM2.5 during the night are largely dominated by sulfate aerosols.

Fig. 10. Spatial distribution differences of monthly-averaged NO2 (μg/m3) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00
LST, c) 14:00 LST, and d) 20:00 LST.

11
T. Chen et al. Applied Geography 120 (2020) 102228

Fig. 11. Spatial distribution differences of monthly-averaged SO2 (μg/m3) simulation results between two land-use map scenarios in January a) 02:00 LST, b) 08:00
LST, c) 14:00 LST, and d) 20:00 LST.

5. Conclusions simulated meteorology changes due to land urbanization over the YRD
region by updating LULC data in the air quality simulation.
This study used two land-use/cover scenarios in the WRF-CMAQ The simulated changes of T2 and PBLH exhibited similar diurnal
model to simulate the impact of land-use/cover changes due to land changes and spatial distributions. T2 and PBLH increased most
urbanization on the meteorology and air quality of the YRD region. We dramatically in the middle of the day. The increase of T2 ranged from
also evaluated the sensitivity of PM2.5, NO2, SO2, and ozone to the 0.2 to 0.7 � C at 14:00 LST whereas PBLH increases by a maximum of

12
T. Chen et al. Applied Geography 120 (2020) 102228

116.74 m at 14:00 LST. As urban areas store more heat, changes in T2 Dirgawati, M., Barnes, R., Wheeler, A. J., Arnold, A. L., Mccaul, K. A., Stuart, A. L., …
Heyworth, J. S. (2015). Development of land use regression models for predicting
and PBLH mainly occur in downtown urban areas. Moreover, land ur­
exposure to NO 2 and NO x in metropolitan perth, western Australia. Environmental
banization significantly changes the surface roughness, leading to Modelling & Software, 74(C), 258–267.
increased surface resistance, which leads to decreased wind speed. The Du, Y., Sun, T., Jian, P., Kai, F., Liu, Y., Yang, Y., et al. (2018). Direct and spillover effects
average reduction of WS10 was 0.29 m/s and the maximum of 0.52 m/s of urbanization on PM2.5 concentrations in China’s top three urban agglomerations.
Journal of Cleaner Production, 190, 72–83.
occurred in the daytime. Engel-Cox, J. A., Holloman, C. H., Coutant, B. W., & Hoff, R. M. (2004). Qualitative and
The modifications of regional meteorological conditions due to land quantitative evaluation of MODIS satellite sensor data for regional and urban scale
urbanization may alter the transport and diffusion of air pollutants and air quality. Atmospheric Environment, 38(16), 2495–2509.
EPA, U. (2007). Guidance on the use of models and other analyses for demonstrating
subsequently air quality. PM2.5 changes exhibited a clear daily cycle and attainment of air quality goals for ozone, PM2.5, and regional haze. US Environmental
urban characteristics. The reduction of PM2.5 was higher during the day Protection Agency, Office of Air Quality Planning and Standards.
than at night, ranging from 1.04 to 6.53 μg/m3, and was more obvious in Fioletov, V. E., Mclinden, C. A., Krotkov, N. A., Moran, M. D., & Yang, K. (2011).
Estimation of SO2 emissions using OMI retrievals. Geophysical Research Letters, 38
urban areas. The PM2.5 concentration decreased in most urban areas (21), 1.
after land urbanization under the same emission source. Changes in the Grell, G. A., & Devenyi, D. (2002). A generalized approach to parameterizing convection
concentration of SO2 ( 4.39–8.51 μg/m3) and NO2 ( 2.25–3.25 μg/ combining ensemble and data assimilation. Geophysical Research Letters, 29(14), 38-
1.
m3), which are precursors of PM2.5, were the main reason for the char­ Habermann, M., Billger, M., & Haeger-Eugensson, M. (2015). Land use regression as
acteristic distribution of PM2.5 changes. SO2 and NO2 changes are also method to model air pollution. Previous results for gothenburg/Sweden. Procedia
related to PBLH and WS10 changes to a certain extent. The change trend Eng, 115, 21–28.
Hoek, G., Beelen, R., Hoogh, K. D., Vienneau, D., Gulliver, J., Fischer, P., et al. (2008).
of O3 concentration was opposite to that of PM2.5 concentration, which
A review of land-use regression models to assess spatial variation of outdoor air
is predominantly affected by changes of NO2 concentration. This also pollution. Atmospheric Environment, 42(33), 7561–7578.
indicates that the air pollution caused by land urbanization will aggra­ Hoff, R. M., & Christopher, S. A. (2009). Remote sensing of particulate pollution from
vate photochemical smog pollution, which is dominated by O3 space: Have we reached the promised land? J. Air. Waste. Manage., 59(6), 642–644.
Ito, T., Ogino, K., Nagaoka, K., & Takemoto, K. (2018). Relationship of particulate matter
(0.5–2.19 μg/m3). and ozone with 3-nitrotyrosine in the atmosphere. Environmental Pollution, 236,
Although this research simulates the impact of land urbanization on 948–952.
regional meteorology and air quality, the results are limited by the Karambelas, A., Holloway, T., Kiesewetter, G., & Heyes, C. (2018). Constraining the
uncertainty in emissions over India with a regional air quality model evaluation.
complex relationship between underlying urban surfaces and air pollu­ Atmospheric Environment, 174, 194–203.
tion. In order to more comprehensively explore the correlation between Liang, Y., Cai, W., & Ma, M. (2019). Carbon dioxide intensity and income level in the
urban surface changes and regional air quality, it is necessary to further Chinese megacities’ residential building sector: Decomposition and decoupling
analyses. The Science of the Total Environment, 677, 315–327.
improve the model parameterization scheme and integrate complex Li, L., Chen, C., Fu, J., Huang, C., Streets, D., Huang, H., … Wang, H. (2011). Air quality
surface parameters such as urban building height information, land-use and emissions in the Yangtze River Delta, China. Atmospheric Chemistry and Physics,
patch shape, and surface thermal environments into the simulation 11(4), 1621–1639.
Lin, Y.-L., Farley, R. D., & Orville, H. D. (1983). Bulk parameterization of the snow field
process. in a cloud model. Journal of Climate and Applied Meteorology, 22(6), 1065–1092.
Overall, our findings can therefore provide important scientific Lin, S., Wang, S., Marinova, D., Zhao, D., & Hong, J. (2017). Impacts of urbanization and
guidance on air quality management to local governments. Moreover, real economic development on CO 2 emissions in non-high income countries:
Empirical research based on the extended STIRPAT model. Journal of Cleaner
this study can help policy makers and urban planners maintain a balance
Production, 166, 952–966.
between land resource utilization and atmospheric environmental Ma, M., Cai, W., Cai, W., & Dong, L. (2019). Whether carbon intensity in the commercial
quality in order to simultaneously achieve economic development and building sector decouples from economic development in the service industry?
ecological protection. Empirical evidence from the top five urban agglomerations in China. Journal of
Cleaner Production, 222, 193–205.
Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., & Clough, S. A. (1997).
Funding Radiative transfer for inhomogeneous atmosphere: RRTM, a validated correlated-k
model for the longwave. Journal of Geophysical Research - D: Atmospheres, 102(D14),
16663–16682.
This work was supported by the National Key R&D Program of China Nenes, A., Pandis, S. N., & Pilinis, C. (1998). Isorropia: A new thermodynamic
(No.2017YFB0504205), Youth Fund Project of Ministry of Education equilibrium model for multiphase multicomponent inorganic aerosols. Aquatic
Geochemistry, 4(1), 123–152.
Humanities and Social Sciences Research (No.16YJCZH161) and China
Noor, N. M., Hasim, N. I. M., & Yusof, S. Y. (2018). Variation of ground-level ozone
Scholarship Council (No. 201706190147). concentration in urbanized area in Malaysia. Materials Science and Engineering, 374
(1), 012097.
Pisoni, E., Albrecht, D., Mara, T. A., Rosati, R., Tarantola, S., & Thunis, P. (2018).
CRediT authorship contribution statement Application of uncertainty and sensitivity analysis to the air quality SHERPA
modelling tool. Atmospheric Environment, 183, 84–93.
Tan Chen: Data curation, Writing - original draft, Writing - review & Pleim, J. E., & Xiu, A. (2003). Development of a land surface model. Part II: Data
assimilation. Journal of Applied Meteorology, 42(12), 1811–1822.
editing. Manchun Li: Conceptualization, Supervision. Lina Luo:
Prakash, K., Jeremiah, J., Greg, Y., & Eladio, K. (2014). Implementation and application
Methodology, Software. Shulin Deng: Investigation, Visualization. Rui of sub-grid-scale plume treatment in the latest version of EPA’s third-generation air
Zhou: Formal analysis, Validation. Dong Chen: Funding acquisition. quality model, CMAQ 5.01. J. Air. Waste. Manage., 64(4), 453–467.
Steinfeld, J. (1998). Atmospheric chemistry and physics: From air pollution to climate
change. Physics Today, 51, 88.
Acknowledgements Su, T., Li, Z., & Kahn, R. (2018). Relationships between the planetary boundary layer
height and surface pollutants derived from lidar observations over China: Regional
pattern and influencing factors. Atmospheric Chemistry and Physics, 18(21),
We thank Quazi Z. Rasool from UNC-Chapel Hill for his invaluable 15921–15935.
help with CMAQ modeling. We also thank Daniel S. Cohan form Rice Wang, J., & Christopher, S. A. (2003). Intercomparison between satellite-derived aerosol
University, who is the sponsor for running models on the DAVinCI optical thickness and PM2.5 mass: Implications for air quality studies. Geophysical
Research Letters, 30(21), 267–283.
cluster system.
Wu, J., Zheng, H., Feng, Z., Xie, W., & Jing, S. (2018). Study on the relationship between
urbanization and Fine Particulate Matter (PM2.5 ) concentration and its implication
References in China. Journal of Cleaner Production, 182, 872–882.
Xie, M., Liao, J., Wang, T., Zhu, K., Zhuang, B., Han, Y., … Li, S. (2016). Modeling of the
anthropogenic heat flux and its effect on regional meteorology and air quality over
Boylan, J. W., & Russell, A. G. (2006). PM and light extinction model performance
the Yangtze River Delta region, China. Atmospheric Chemistry and Physics, 16(10),
metrics, goals, and criteria for three-dimensional air quality models. Atmospheric
6071–6089.
Environment, 40(26), 4946–4959.
Xu, Q., Dong, Y. X., & Yang, R. (2018). Urbanization impact on carbon emissions in the
Chen, X., Liu, Y., Lai, A., Han, S., Fan, Q., Wang, X., & Fan, S. (2018). Factors dominating
Pearl River Delta region: Kuznets curve relationships. Journal of Cleaner Production,
3-dimensional ozone distribution during high tropospheric ozone period. Environ.
180, 514–523.
Pollut., 232, 55–64.

13
T. Chen et al. Applied Geography 120 (2020) 102228

Xu, H., & Zhang, W. (2016). The causal relationship between carbon emissions and land Zhang, J. J. Y., Sun, L., Barrett, O., Bertazzon, S., Underwood, F. E., & Johnson, M.
urbanization quality: A panel data analysis for Chinese provinces. Journal of Cleaner (2015). Development of land-use regression models for metals associated with
Production, 137, 241–248. airborne particulate matter in a North American city. Atmospheric Environment, 106,
Yang, X., Wang, S., Zhang, W., Zhan, D., & Li, J. (2016). The impact of anthropogenic 165–177.
emissions and meteorological conditions on the spatial variation of ambient SO 2 Zhang, G., Zhang, N., & Liao, W. (2018). How do population and land urbanization affect
concentrations: A panel study of 113 Chinese cities. The Science of the Total CO2 emissions under gravity center change? A spatial econometric analysis. Journal
Environment, 584–585, 318. of Cleaner Production, 202, 510–523.
Yarwood, G., Rao, S., Yocke, M., & Whitten, G. (2005). Updates to the carbon bond Zheng, B., Zhang, Q., Zhang, Y., He, K., Wang, K., Zheng, G., … Kimoto, T. (2015).
chemical mechanism: CB05. Final Report prepared for US EPA. Website www.camx. Heterogeneous chemistry: A mechanism missing in current models to explain
com/publ/pdfs/CB05_Final_Report_120805. pdf. secondary inorganic aerosol formation during the january 2014 haze episode in
Zhang, N., Gao, Z., Wang, X., & Chen, Y. (2010). Modeling the impact of urbanization on north China. Atmospheric Chemistry and Physics, 14, 16731–16776.
the local and regional climate in Yangtze River Delta, China. Theoretical and Applied
Climatology, 102(3–4), 331–342.

14